Use RecursivelyDeleteDeadPHINode instead of cleaning IndVar manually

Author: mshockwave

llvm/lib/CodeGen/EVLIndVarSimplify.cpp

@@ -27,6 +27,7 @@
 #include "llvm/Support/MathExtras.h"
 #include "llvm/Support/raw_ostream.h"
 #include "llvm/Transforms/Scalar/LoopPassManager.h"
+#include "llvm/Transforms/Utils/Local.h"
 
 #define DEBUG_TYPE "evl-iv-simplify"
 
@@ -105,28 +106,6 @@ static uint32_t getVFFromIndVar(const SCEV *Step, const Function &F) {
   return 0U;
 }
 
-// Remove the original induction variable if it's not used anywhere.
-static void tryCleanupOriginalIndVar(PHINode *OrigIndVar,
-                                     const InductionDescriptor &IVD) {
-  if (OrigIndVar->getNumIncomingValues() != 2)
-    return;
-  Value *InitValue = OrigIndVar->getIncomingValue(0);
-  Value *RecValue = OrigIndVar->getIncomingValue(1);
-  if (InitValue != IVD.getStartValue())
-    std::swap(InitValue, RecValue);
-
-  // If the only user of OrigIndVar is the one that produces RecValue, then we
-  // can safely remove it.
-  if (!OrigIndVar->hasOneUse() || OrigIndVar->user_back() != RecValue)
-    return;
-
-  LLVM_DEBUG(dbgs() << "Removed the original IndVar " << *OrigIndVar << "\n");
-  // Turn OrigIndVar into dead code by replacing all its uses by the initial
-  // value of this loop.
-  OrigIndVar->replaceAllUsesWith(InitValue);
-  OrigIndVar->eraseFromParent();
-}
-
 bool EVLIndVarSimplifyImpl::run(Loop &L) {
   if (!EnableEVLIndVarSimplify)
     return false;
@@ -262,7 +241,14 @@ bool EVLIndVarSimplifyImpl::run(Loop &L) {
   auto *NewPred = Builder.CreateICmp(Pred, EVLIndVar, TC);
   OrigLatchCmp->replaceAllUsesWith(NewPred);
 
-  tryCleanupOriginalIndVar(IndVar, IVD);
+  // llvm::RecursivelyDeleteDeadPHINode only deletes cycles whose values are
+  // not used outside the cycles. However, in this case the now-RAUW-ed
+  // OrigLatchCmp will be consied a use outside the cycle while in reality it's
+  // practically dead. Thus we need to remove it before calling
+  // RecursivelyDeleteDeadPHINode.
+  (void)RecursivelyDeleteTriviallyDeadInstructions(OrigLatchCmp);
+  if (llvm::RecursivelyDeleteDeadPHINode(IndVar))
+    LLVM_DEBUG(dbgs() << "Removed original IndVar\n");
 
   ++NumEliminatedCanonicalIV;
 

llvm/test/CodeGen/RISCV/evl-iv-simplify.ll

@@ -20,8 +20,6 @@ define void @simple(ptr noalias %a, ptr noalias %b, <vscale x 4 x i32> %c, i64 %
 ; CHECK-NEXT:    [[N_RND_UP:%.*]] = add i64 [[N]], [[TMP8]]
 ; CHECK-NEXT:    [[N_MOD_VF:%.*]] = urem i64 [[N_RND_UP]], [[TMP5]]
 ; CHECK-NEXT:    [[N_VEC:%.*]] = sub i64 [[N_RND_UP]], [[N_MOD_VF]]
-; CHECK-NEXT:    [[TMP9:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT:    [[TMP10:%.*]] = mul i64 [[TMP9]], 4
 ; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
 ; CHECK:       vector.body:
 ; CHECK-NEXT:    [[EVL_BASED_IV:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_EVL_NEXT:%.*]], [[VECTOR_BODY]] ]
@@ -37,9 +35,7 @@ define void @simple(ptr noalias %a, ptr noalias %b, <vscale x 4 x i32> %c, i64 %
 ; CHECK-NEXT:    call void @llvm.vp.store.nxv4i32.p0(<vscale x 4 x i32> [[TMP18]], ptr align 4 [[TMP20]], <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer), i32 [[TMP12]])
 ; CHECK-NEXT:    [[TMP21:%.*]] = zext i32 [[TMP12]] to i64
 ; CHECK-NEXT:    [[INDEX_EVL_NEXT]] = add i64 [[TMP21]], [[EVL_BASED_IV]]
-; CHECK-NEXT:    [[INDEX_NEXT:%.*]] = add i64 0, [[TMP10]]
 ; CHECK-NEXT:    [[TMP23:%.*]] = icmp eq i64 [[INDEX_EVL_NEXT]], [[N]]
-; CHECK-NEXT:    [[TMP26:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]]
 ; CHECK-NEXT:    br i1 [[TMP23]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]]
 ; CHECK:       middle.block:
 ; CHECK-NEXT:    br i1 true, label [[FOR_COND_CLEANUP:%.*]], label [[SCALAR_PH]]
@@ -166,8 +162,6 @@ define void @fixed_iv_step(ptr %arg0, ptr %arg1, i64 %N) #0 {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    br label [[VECTOR_PH:%.*]]
 ; CHECK:       vector.ph:
-; CHECK-NEXT:    [[N_RND_UP:%.*]] = add nsw i64 [[N]], 15
-; CHECK-NEXT:    [[N_VEC:%.*]] = and i64 [[N_RND_UP]], -16
 ; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 2 x ptr> poison, ptr [[ARG0]], i64 0
 ; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 2 x ptr> [[BROADCAST_SPLATINSERT]], <vscale x 2 x ptr> poison, <vscale x 2 x i32> zeroinitializer
 ; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
@@ -179,9 +173,7 @@ define void @fixed_iv_step(ptr %arg0, ptr %arg1, i64 %N) #0 {
 ; CHECK-NEXT:    tail call void @llvm.vp.store.nxv2p0.p0(<vscale x 2 x ptr> [[BROADCAST_SPLAT]], ptr align 8 [[GEP]], <vscale x 2 x i1> shufflevector (<vscale x 2 x i1> insertelement (<vscale x 2 x i1> poison, i1 true, i64 0), <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer), i32 [[TMP1]])
 ; CHECK-NEXT:    [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
 ; CHECK-NEXT:    [[INDEX_EVL_NEXT]] = add i64 [[EVL_BASED_IV]], [[TMP2]]
-; CHECK-NEXT:    [[LSR_IV_NEXT33:%.*]] = add i64 [[N_VEC]], -16
 ; CHECK-NEXT:    [[TMP4:%.*]] = icmp eq i64 [[INDEX_EVL_NEXT]], [[N]]
-; CHECK-NEXT:    [[TMP5:%.*]] = icmp eq i64 [[LSR_IV_NEXT33]], 0
 ; CHECK-NEXT:    br i1 [[TMP4]], label [[FOR_END_LOOPEXIT5:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP3]]
 ; CHECK:       for.end.loopexit5:
 ; CHECK-NEXT:    br label [[FOR_END:%.*]]
@@ -244,8 +236,6 @@ define void @fixed_iv_step_tc(ptr %arg0, ptr %arg1) #0 {
 ; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    br label [[VECTOR_PH:%.*]]
 ; CHECK:       vector.ph:
-; CHECK-NEXT:    [[N_RND_UP:%.*]] = add nsw i64 87, 15
-; CHECK-NEXT:    [[N_VEC:%.*]] = and i64 [[N_RND_UP]], -16
 ; CHECK-NEXT:    [[BROADCAST_SPLATINSERT:%.*]] = insertelement <vscale x 2 x ptr> poison, ptr [[ARG0]], i64 0
 ; CHECK-NEXT:    [[BROADCAST_SPLAT:%.*]] = shufflevector <vscale x 2 x ptr> [[BROADCAST_SPLATINSERT]], <vscale x 2 x ptr> poison, <vscale x 2 x i32> zeroinitializer
 ; CHECK-NEXT:    br label [[VECTOR_BODY:%.*]]
@@ -257,9 +247,7 @@ define void @fixed_iv_step_tc(ptr %arg0, ptr %arg1) #0 {
 ; CHECK-NEXT:    tail call void @llvm.vp.store.nxv2p0.p0(<vscale x 2 x ptr> [[BROADCAST_SPLAT]], ptr align 8 [[GEP]], <vscale x 2 x i1> shufflevector (<vscale x 2 x i1> insertelement (<vscale x 2 x i1> poison, i1 true, i64 0), <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer), i32 [[TMP1]])
 ; CHECK-NEXT:    [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
 ; CHECK-NEXT:    [[INDEX_EVL_NEXT]] = add i64 [[EVL_BASED_IV]], [[TMP2]]
-; CHECK-NEXT:    [[LSR_IV_NEXT33:%.*]] = add i64 [[N_VEC]], -16
 ; CHECK-NEXT:    [[TMP3:%.*]] = icmp eq i64 [[INDEX_EVL_NEXT]], 87
-; CHECK-NEXT:    [[TMP4:%.*]] = icmp eq i64 [[LSR_IV_NEXT33]], 0
 ; CHECK-NEXT:    br i1 [[TMP3]], label [[FOR_END_LOOPEXIT5:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP3]]
 ; CHECK:       for.end.loopexit5:
 ; CHECK-NEXT:    br label [[FOR_END:%.*]]